When treating a colon adenocarcinoma (CA) patient, after surgical resection of the tumor the clinician must formulate a plan for adjuvant systemic therapy. This decision is based upon an assessment of the risk of systemic disease recurrence, and is currently informed by pathological factors such as stage, histological grade, and lymph node status. Improvement of the accuracy of risk assessment for individual patients is an area of recognized need. Much of the current information used to assess risk focuses on the cells within tumors, including their morphological properties. Less attention is paid to the extracellular matrix through which metastasizing cells must travel. Second harmonic generation (SHG) is an optical scattering phenomenon whose directionality (as quantified by the ?F/B? ratio) is affected by the diameter, spacing, and disorder of fibrils within collagen fibers. Our preliminary data suggests that F/B analysis of tumor samples provides prognostic information about future metastasis that is ?matrix-focused? and hence complementary to current ?cell-focused? methods. In a preliminary study in 44 Stage I colon adenocarcinoma samples we found that F/B of the primary tumor is a significant prognostic indicator of progression free survival time. Significantly, the quartile of Stage I patients with the lowest F/B ratio had a clinical outcome indistinguishable from Stage III patients: a 15 year progression free survival percentage of below 50%. In other words, in this study F/B identified a subset of Stage I patients who had survival statistics similar to Stage III patients. Stage I patients are rarely prescribed adjuvant chemotherapy while Stage III patients are almost always prescribed it. This suggests that F/B might be able to identify individual patients who would benefit from adjuvant chemotherapy and who would be left untreated based upon current prognostic indicators. The prognostic trend was also evident in a cohort of 72 Stage II colon adenocarcinoma samples, although it was not significant. Consequently we hypothesize that F/B is a clinically useful predictor of metastatic outcome in colon adenocarcinoma. This project will move this idea closer to the clinic by first (Aim 1) using archived samples and follow up data in separate training and validation sets to develop and test predictive algorithms that include F/B, in addition to clinical and genomic information. Second it will (Aim 2) quantify the effect of adjuvant chemotherapy on the predictive ability of the algorithms, as well as quantify their ability to predict chemotherapeutic efficacy. We predict that F/B analysis will be an effective tool that can reach the clinic rapidly after this study to improve metastatic risk assessment. Improving the accuracy of risk estimation for an individual patient will allow clinicians to treat those patients who are destined for metastases, improving outcomes, while avoiding treatment for those patients who are not, reducing overtreatment.!